Background: The current (seventh edition) American Joint Commission on Cancer (AJCC) Staging System for pancreatic ductal adenocarcinoma (PDAC) dichotomizes pathologic lymph node (LN) involvement into absence (pN0) or presence (pN1) of disease. The recently announced eighth edition also includes stratification on the number of positive nodes. Furthermore, LNs detected on preoperative imaging (CT, MRI, or endoscopic ultrasound-EUS) are considered to be pathologically involved in other gastrointestinal cancers. However, this is less well defined for PDAC. Therefore, the three aims of this study were to determine (1) whether the new AJCC staging system led to more accurate staging, (2) the number of nodes needed to be examined to detect pathologic involvement, and (3) if pN disease could be reliably detected on preoperative imaging in PDAC.
Methods: A retrospective review of all patients undergoing pancreatectomy at a single US academic center from January 1990 to September 2015. Pathology reports of resected specimens were reviewed to determine the total number of LNs examined and those positive for metastasis. CT, MRI, and/or EUS reports were used to determine the presence or absence of preoperatively detectable LN enlargement.
Results: Of the 490 surgical resections for PDAC, pN1 disease was detected in 59.4% (n = 291) and was positively correlated with the number of LNs pathologically examined (P < 0.001). Patients with pN1 disease had a shorter overall survival (OS) than those without nodal involvement (25.1 vs. 44.0 months; P < 0.001); however, OS was not different when stratifying by the number of nodes as on the eighth AJCC system. Pathologic examination of > 20 LNs in treatment naïve patients was optimal to detect pN1 disease and predict longer OS for those without nodal involvement (median survival > 41.1 months, P = 0.03 when compared to < 15 or 15-19 LNs examined). LNs were detected by CT, MRI, or EUS in 30.7% (103/335) of patients. The positive predictive value (PPV) of preoperative LN detection for pathologic involvement was 77.3% for treatment naïve patients and 84.2% for those without biliary obstruction.
Conclusions: Although the LN scoring in the seventh PDAC AJCC Staging System was sufficient to predict OS of our patients, more LNs than previously considered (20 vs. 15) were optimal to detect pathologic involvement. Preoperative LN detection was an accurate predictor of pN1 disease for treatment naïve patients without biliary obstruction.
Keywords: Biliary obstruction; Lymph node; Pancreatic ductal adenocarcinoma.